Cross Border Carve-out Initial Returns and Long-term Performance

Financial Decisions, Winter 2012, Article 3 Abstract Cross Border Carve-out Initial Returns and Long-term Performance Thomas H. Thompson Lamar University This study examines initial period and three-year returns for cross border carve-outs from 1988 to 2006 and finds several factors that maintain their significance over a three-year period after equity carve-out offerings. We report several differences between cross border carve-outs and carve-outs with US traded parents. Also, we observe that our independent variables explain approximately 20% of the multiple regression initial returns for cross border carve-outs. Moreover, we report that, contrary to Vijh (1999), negative three-year carve-out returns are statistically significant. JEL classification: G32, G34 Keywords: Carve-outs, Spin-offs, Divestitures

1. Introduction US companies have used equity carve-outs as a reorganization tool for some time. We investigate the 76 equity carve-outs foreign parents initiated for their US subsidiaries during the period from 1988 to 2006. Although there have been other long-term carve-out and IPO studies, this is the first study to examine the returns of cross border carve-outs (US traded carve-outs with foreign parents). The results of the partial liquidation of their US traded subsidiaries (cross border carve-outs) show that foreign parents can take advantage of the larger offering sizes and lower initial returns (underpricing). We observe several significant differences between carve-outs with US traded parents and crossborder carve-outs. First, although offering proceeds are higher for cross border carve-outs, initial returns are 10.61% lower than for carve-outs with US traded parents. Second, US traded carveout parents retain a higher percentage of ownership. This leads to higher overhang. Third, cross border carve-outs have higher percentages of secondary offerings. In addition, for cross border carve-out multiple regressions, we observe that our independent variables explain 20.16% of the initial returns. Also, we observe higher three-year returns for cross border carve-outs than for equity carve-outs or for IPOs during the same period. Moreover, we observe, contrary to Vijh (1999), that three year carve-out returns are negative and statistically significant. The mean IPO offering from 1988 to 2006 is $68.5 million (Ritter, 2007). As the initial leg of a two-stage process and with approximately 34% of the subsidiary s value offered, mean offering sizes for carve-outs with foreign parents are $339.01 million, over four times the mean IPO offering for a similar period. The large carve-out offering sizes imply low information asymmetry and lower underpricing than for regular IPOs (Vijh, 1999; and Carter, Dark, and Singh, 1998). 2. Background Schipper and Smith (1986) note that subsidiaries benefit from carve-outs in many ways. First, the subsidiary can obtain separate financing for its own growth opportunities. Next, the newly traded company provides increased information disclosure and thus reduces uncertainty. Also, subsidiary managers can be motivated to increase shareholder wealth and their own as well, if awarded shares through stock ownership plans. Carve-outs imply lower risk than for typical IPOs due to larger offering size (Vijh, 1999) and reduced information asymmetry because the former subsidiaries were reported as part of existing firms (Schipper and Smith, 1986). Thus, underwriters need not underprice (discount) carve-out prices. Ritter (1991) and Loughran and Ritter (1995) show that IPOs, in general, have high initial returns, but negative three-year returns. For carve-outs, Vijh (1999, 2002) reports lower subsidiary initial period returns and higher long-term returns than for IPOs. We observe that cross border carve-outs have lower initial returns and higher long-term returns than for all US equity carve-outs. This reflects lower risk for cross border carve-outs. The carve-out initial public offering generates cash that can be used to retire debt, to increase cash for investment, or to cash-out insiders. Also, carve-outs allow parents to showcase their subsidiaries to prospective buyers (Klein, Rosenfeld, and Beranek, 1991). In addition, the carve- 1

out phase of the two-stage combination creates a market for the new subsidiary s stock prior to a spin-off and provides more corporate information from additional analyst coverage (Schipper and Smith, 1986). Low levels of secondary offerings -- reported insider selling of the carve-out IPO -- reflect insiders preferences hold the shares to be spun-off and reap the potential market price increases (Leland and Pyle, 1977; and Loughran and Ritter, 2002). This creates an overhang, the ratio of shares retained by the issuer or insiders for future resale divided by the shares offered. Also, we observe a higher level of secondary offerings in cross-border carve-outs and no second stage spin-offs. This indicates that parent companies seek to increase proceeds from the initial offering instead of waiting for a subsequent offering or sale of parent controlled shares. 3. Sample selection and data To be included in our sample an equity carve-out must have an initial offering price of five dollars or more and be traded on the NYSE, Amex or Nasdaq (over-the-counter). Also, the parent must retain one-third or more ownership in the subsidiary after the carve-out. For the 1990 and prior period we examined new releases for initial public offerings of subsidiaries. For the post-1990 period we started with the list of carve-outs in Mergers and Acquisition Magazine. With CRSP data, we identify 446 carve-outs from 1988 to 2006. We eliminate 6 companies with joint foreign and US parents. We identify 76 companies with foreign parents and contrast these cross border carve-outs with the 364 carve-outs with US traded parents. 1 All carve-outs are cross checked with Lexis-Nexis, the Investment Dealer s Digest, Mergent Industrial Manual (formerly Moody s) and the Dow Jones News Service. SEC filings and news articles provide additional information for the sample. The Center for Research in Security Prices (CRSP) provides stock return data. Consistent with Bradley and Jordan (2002) and Loughran and Ritter (2002), the carve-out subsidiary initial return is the holding period return from the offering price to first-day closing price. Three-year returns are annualized time-weighted weekly returns. Similar to Ritter (1991) and Hulbert, Miles, and Woolridge (2002) we subtract CRSP equalweighted index results (to include distributions) from carve-outs stock returns to obtain marketadjusted returns. 2 4A. Variable definition 4. Empirical analysis and initial results In the following paragraphs we define our variables and provide their economic intuition. Using these factors we construct our multivariate cross-sectional tests for stock returns. Mulherin and Boone (2000) report that the log of offering size (LSIZ) can be positively related to stock returns and reflect a wealth increase. However, Asquith and Mullins (1986) and Mikkelson and Partch (1986) observe a negative correlation between size and stock returns. We use LSIZ to capture the size effect. 1 Since only eleven cross border carve-outs were conducted during the bubble period (1999-2000) we examine only full period returns and avoid the problems with a small sample size. 2 We use the CRSP equal-weighted index to avoid small company bias (Ritter, 1991). 2

Loughran and Ritter (2004) show that announcement period returns can vary positively with the percentage of ownership retained (RET). This percentage will be given up by the parent in a divestiture. Bradley and Jordan (2002) show that overhang, the ratio of subsidiary shares retained to shares sold in the offering, provides a positive signal for investors, and later benefits from the subsidiary s price appreciation. Carve-out companies sell no more than two-thirds of their ownership at initial offerings. Thus, overhang, unsold subsidiary shares divided by the shares sold, may be related to underpricing, as defined as the percentage change in price from the initial offering price to the first closing price. However, only the shares from the offering are underpriced. The subsidiary shares retained by parent company insiders reflect their market value. Thus, for a given level of underpricing, share dilution declines as overhang increases (Bradley and Jordan, 2002). Using the percentage of the subsidiary shares retained by the parent (RET) we define overhang as: Overhang = RET/ (1-RET) (1) Loughran and Ritter (2002), in their prospect theory, show that insiders receive a wealth increase despite the underpricing caused dilution of share values. If the offering is small relative to the pre-issue shares outstanding, the wealth increase for non-selling shareholders will exceed the dollar loss due to underpricing and dilution. Thus, only the shares sold are underpriced. Given equity carve-out parents retain at least a one-third ownership (with a mean of 70%), overhang can be 0.50 or greater. Sarbanes-Oxley and the implementing SEC regulations require that a parent must retain a 50% or greater ownership to retain control of a subsidiary. Thus, parents will retain a majority ownership and the overhang is one or greater. Loughran and Ritter (2002) suggest two reasons for overhang. First, based on supply and demand, there may be a scarcity premium for firms that sell a small fraction of the company to the public. Thus, it is most likely that equity carve-outs will generate wealth increases for nonselling shareholders. A second reason is that there may be an optimal amount of capital to be raised in an IPO. Generally, the shares retained by carve-out parents are authorized, but unissued. In their managerial discretion hypothesis, Allen and McConnell (1998) found that carve-out returns vary with the proportion of proceeds used to retire subsidiary debt or to pay dividends from the total offering proceeds (DEBT). The USE coefficient indicates the proportion of proceeds reinvested by the issuer. Mikkelson and Partch (1986) report that IPO returns vary with the percentage of offering funds used by issuing firms. Also, Schipper and Smith (1986) note that companies that use carve-out funds for positive net present value projects can have high returns. In their managerial discretion hypothesis Allen and McConnell (1998), contrary to Mikkelson and Partch (1986), show positive returns for carve-out firms that pay down debt or pay dividends with their proceeds. They offer that corporations initiate carve-outs due to their high levels of leverage or operational troubles. Thus, stock returns for carve-outs that use the funds to reduce debt or pay dividends to shareholders have higher returns than companies that retain the funds. Thus, the USE factors that vary 3

inversely with (vary with) returns indicate support (contradiction) for the managerial discretion hypothesis. The fraction of the issue that is a secondary offering (SEC) is the ratio of insider shares sold to the total shares offered. Logue (1973) offers that secondary issuers, due to their strong bargaining position, can demand lower discounts (higher prices). Thus, initial returns should vary inversely with the SEC coefficient. The CBOE Volatility Index (VXO) represents the implied volatility of S&P 100 Index options. The VXO predicts future index returns and represents market ebullience and the potential for future public offerings (Dennis, Mayhew, and Stivers, 2006). A high (low) VXO indicates high (low) risk for public offerings. Also, the VXO variable provides a proxy for market ebullience as defined by Logue (1973). Expanding Hanley s (1993) study, Bradley and Jordan (2002) and Loughran and Ritter (2002) show that the timing of the increase or decrease of the filing range influences the underpricing. Bradley and Jordan (2002) use the following measures for increases and decreases of initial filing ranges: UP1 (DW1) is the percentage difference between the original midrange file price and the amended midrange file price for those companies that amend their file range up (down), zero otherwise. The variables are defined as: UP1 = max [0, (midpoint amended - midpoint original)/midpoint original] * 100% (2) DW1 = min [0, (midpoint amended - midpoint original)/midpoint original] * 100% (3) where amended and original reflect the original and final amended file ranges. These two variables allow for the possibility of an asymmetric effect from file range increases or decreases. Next, two variables capture possible asymmetric effects between the final file range and the final offering price. UP2 (DW2) is the percentage difference between the final midrange file price and the final offer price for those companies that have offer prices above (below) the amended midrange file price, zero otherwise. Loughran and Ritter (2002) advance a leaning against the wind hypothesis that provides an alternative explanation to partial price adjustment. They offer that IPO investors tend to overreact and bid the price above the long-term value. Then, IPO prices revert to their long-term values. Thus, a testable implication of this hypothesis is that there should be a negative correlation between first day returns and subsequent long-term returns. Prestigious underwriters can help their clients successfully complete the process and maximize wealth. Loughran and Ritter (2004) revise the Carter and Manaster (1990) investment banker reputation rankings that are based on the underwriters relative order on the prospectus (tombstone). The Loughran and Ritter (2004) ratings range from a low of 1.0 to a high of 9.1 and account for penny stock IPOs that are avoided by prestigious investment bankers. In addition, Beatty and Welch (1996) show that underwriter reputation varies with returns after 1990 and indicate that investment bankers have redeemed their reputation, took more risk in order to avoid the loss of underwriting opportunities such as the possibility of missing a hot IPO issue. 4

Table 1: Summary Statistics This table shows the annual frequency of equity carve-outs during the period 1988 through 2006, the mean and median initial returns (offering price to first day close), three-year returns, fraction of the subsidiary s outstanding shares retained by parent firms following carve-outs, the mean and median gross proceeds raised in the carve-out, and the offering proportion used for debt reduction or to pay dividends, offering proportion reinvested by the company, the percentage of the offering as a secondary offering, overhang (the ratio of shares retained by the parent to the shares offered), the survivability in years, and a control variable for the Sarbanes-Oxley Act of 2002. Returns are adjusted by concurrent CRSP equal weighted returns. Column 1 reports results for subsidiaries with US parents. Column 2 reports results for cross border carve-outs (US traded subsidiaries with foreign parents). Column 3 shows the difference between factors for carve-outs with US traded parents (Column 1) and cross border carve-outs (Column 2). Column 4 provides t-statistics for the differences. (1) (2) (3) (4) Carve-outs Cross border Differences between with US parents carve-outs Columns 1 and 2 t-stats N 364 76 Initial market-adjusted returns 19.55% 8.93% 10.61% 3.53*** (7.12%) (5.16%) Three-year market-adjusted returns -0.78% -0.62% -0.16% -1.65 * (-0.61%) (-0.52%) Fraction of shares retained by parent 0.72 0.65 0.07 3.30*** (0.75) (0.66) Investment banker reputation 8.28 8.32-0.03-0.20 (9.10) (9.10) Offering proceeds ($millions) 281.07 339.01-57.93-0.76 (85.07) (115.25) Offering proportion used for debt reduction 36.57% 31.37% 5.19% 1.06 (18.50%) (4.00%) Offering proportion reinvested by the company 46.32% 42.39% 3.93% 0.75 (40.00%) (37.00%) Offering proportion as secondary offering 17.12% 25.85% -8.72% -1.86 ** (0.00%) (0.00%) Overhang 5.13 2.72 2.41 4.29*** (3.00) (1.94) Survivability (years) 2.77 2.84-0.07-1.19 (3.00) (3.00) SOX 0.34 0.18 0.16 3.09*** 0.00 0.00 Significance at the 10% and 1% levels is indicated by 1 and 3 asterisks, respectively. 5

4B. Descriptive statistics Table 1 provides summary statistics. Columns 1 and 2 provide results for carve-outs with US traded parents and cross border carve-outs for the period from 1988 to 2006, respectively. Both columns show mean and median for the fraction of the subsidiary s ownership retained by the parent, offering proceeds, and the offering proportion used for debt reduction or to pay dividends. There are 440 carve-outs from 1988 to 2006. The percent of subsidiary ownership retained ranges from one-third to 93 % with a mean of 70% and a median of 74%. Column 3 shows the difference between factors for carve-outs with US traded parents (Column 1) and cross-border carve-outs (Column 2). Column 4 provides t-values for the differences. For the 76 cross border carve-outs, the percent of ownership varies from 33% to 96% with a mean (median) of 65% (66%). These results are similar to the Allen and McConnell (1998) mean (median) of 69% (80%) and the Vijh (2002) median of 71.9%. Carve-outs with US traded parents the mean (median) gross offering proceeds and marketadjusted initial returns are $281.07 million ($85.07 million) and 19.55% (7.12%), respectively. Cross border carve-outs have mean (median) offering proceeds of $339.01 million ($115.25 million) and have mean (median) market-adjusted initial returns are 8.93% (5.16%). The larger offering sizes and lower initial returns indicate lower risk for carve-outs with foreign parents. Pure IPOs for the same period have a mean offering size of $72.83 million (Ritter, 2007). Carveout mean offering sizes are four times greater than the average IPO. Given that approximately 20% of the shares of subsidiaries are sold in the carve-out offering, the market capitalization of these partial IPOs can be five times the offering size. Thus, the percentage retained by the parent and subsidiary offering sizes are significant. We report -0.78% (-0.61%) and -0.62% (-0.52%) mean (median) three-year market-adjusted returns for carve-outs with US parent and for cross border carve-outs, respectively. Vijh (1999) reports -1.26% three-year market-adjusted returns for all carve-outs. We observe, but do not report in the tables, that ten parents are from the United Kingdom and eight parent companies each are from Canada and Japan. All other countries have five or less parent companies. Thus, no country dominates the sample. Also, almost all parents are from developed countries. We explain significant differences between the variables for carve-outs with US traded parents (Table 1, column 1) and cross-border carve-outs (Table 1, column 2). below. The differences between carve-outs with US traded parents and cross border carve-outs in initial market returns (10.61%), shares retained by parents (0.07), overhang (2.41), and the SOX variable (0.16) are significant at the 1% level. Supporting the prospect theory, greater share retention and overhang lead to higher initial returns for carve-outs with US traded parents. For the post Sarbanes-Oxley period, the carve-outs with US parents subset is 34% of the total versus 18% for cross border carve-outs. The cross border carve-outs have 8.72% higher percentage of secondary offerings (sales by insiders), significant at the 5% level. This reflects that parents of cross border carve-outs prefer to reap more proceeds at the initial offering. The -0.16% difference in three-year returns is marginally significant. 6

Table 2: Single Factor Analysis Initial Returns This table shows market adjusted initial returns for the period from 1988 to 2006 for subsidiaries with US parents (column 1); cross border carve-outs (US traded subsidiaries with foreign parents) (column 2); and t-statistics for differences between variables (column 3). Returns are adjusted by subtracting the contemporaneous returns for the CRSP equal-weighted index. The market-adjusted returns are regressed separately on several independent variables: Overhang (OVER), RET/(1-RET); the log of offering size (LSIZ); the proportion of the offering proceeds reinvested by the company (USE); the proportion of the offering used to pay dividends or to pay down debt (DEBT); the proportion of the offering as a secondary offering (SEC); the CBOE Volatility Index (VXO); and the ratio of the mid-point of the adjusted filing range to the initial filing range where UP1 (DW1) is positive (negative) and zero otherwise; the ratio of the offering price to the mid-point of the final filing range where UP2 (DW2) is an increase (decrease) and zero otherwise; and a control variable for the Sarbanes-Oxley Act where 1 is a 2002 or later carve-out and 0 otherwise. (t-values are in parentheses) (1) (2) (3) T-statistics for Carve-outs withus parents Cross border carve-outs Differences N 364 76 Mean 0.1955 0.0893 3.53 *** [Median] [0.0712] [0.0516] OVER 0.0006-0.0004 4.29 *** (0.23) (-0.06) LSIZ 0.0022-0.0042-0.76 (0.13) (-0.33) USE 0.1960 0.1171 0.75 (3.48)*** (2.78) *** DEBT -0.1086-0.1164 1.06 (-1.88)* (-2.55) ** SEC -0.1470-0.0181-1.86 ** (-2.03)** (-0.38) VXO 1.4825 0.6273 1.02 (3.71)*** (2.08) ** UP1 1.5128 0.1536 1.66 ** (9.79)*** (0.73) DW1 0.7876 0.5019-1.54 * (2.56)** (1.65) UP2 2.3205 0.4050 0.40 (8.04)*** (1.80) * DW2 1.4051 0.4120 1.51 * (4.63) *** (2.05) ** SOX -0.0015-0.0088 3.09 *** (-0.03) (-0.19) Significance at the 10%, 5%, and 1% levels is indicated by 1, 2, and 3 asterisks, respectively. 7

4C. Single factor regressions - initial returns Table 2 reports the single factor regressions of initial returns against several factors that can influence stock returns for stockholders. The factors detailed below are statistically significant and maintain their expected coefficient sign. The positive USE, VXO, and DW2 coefficients are statistically significant at the 5% level or higher for carve-outs with US parents and cross border carve-outs. The positive UP2 parameter is significant at the 1% level for carve-outs with US parents, but marginally significant for cross border carve-outs. For carve-outs with US parents (Table 2, column 1) the negative SEC and positive UP1 and DW1 parameters are significant at the 5% level or higher. The UP1, DW1, UP2, and DW2 results indicate support for the Bradley and Jordan (2002) partial price adjustment hypothesis. The USE results indicate support for Mikkelson and Partch (1986). The negative DEBT coefficient is significant at the 10% and 5% levels for carve-outs with US parents and cross border carve-outs, respectively. These results contradict the managerial discretion hypothesis of Allen and McConnell (1998) and supports Mikkelson and Partch (1986). The positive VXO parameter indicates that returns vary with increased risk. Consistent with Logue (1973), the SEC results indicate that returns vary negatively with the percent of the offering that is a secondary offering. Table 2, column 3 reports t-statistics for differences between variables for carve-outs with US parents and for cross border carve-outs. Carve-outs with US parents have initial returns 10.61% greater than those for cross border carve-outs. In addition to the differences reported in Table 1, the positive difference in the UP1 variables is significant at the 5% level. The differences in the negative DW1 and positive DW2 variables are marginally significant. 4D. Multivariate analysis initial returns Prior to performing our multiple regression analysis, we use the variance inflation factor (VIF) to check our significant variables for multicollinearity in both periods. Since the USE, DEBT, and SEC parameters are highly correlated we delete the SEC variable because it is not significant for cross border carve-outs, the subject of this study. 3 After this adjustment, we observe that all VIF are close to 1. This indicates that there is no further threat of multicollinearity. Thereafter, we apply the Akaike Information Criterion (AIC) to the remaining significant variables from the single variable analysis (Section 4.3) to obtain the optimal variables for the multiple regressions. Similar to the adjusted R 2, the AIC provides a tradeoff between maximizing the explained variance, and limiting an increase in the number of independent variables. Initially we select the models with the lowest AIC for each period. Thereafter we use all variables from the two models to contrast the significance of coefficients over the two periods. The explanatory power for the resulting models approaches that for the lowest AIC models. Table 3 reports multiple regression results for initial returns. 3 The AIC process eliminates the SOX variable for the initial returns multiple regression. 8

Table 3: Multiple Regression Results Initial Returns This table shows multiple regressions of market adjusted initial returns for the total period (1988-2006) for subsidiaries with US parents (column 1)and cross border carve-outs (US traded subsidiaries with foreign parents) (column 2). Returns are adjusted by subtracting the contemporaneous returns for the CRSP equal-weighted index. The market-adjusted returns are regressed on several independent variables: Overhang (OVER), RET/(1-RET); the log of offering size (LSIZ); the proportion of the offering proceeds reinvested by the company (USE); the proportion of the offering used to pay dividends or to pay down debt (DEBT); the CBOE Volatility Index (VXO); and the ratio of the mid-point of the adjusted filing range to the initial filing range where UP1 (DW1) is positive (negative) and zero otherwise; the ratio of the offering price to the mid-point of the final filing range where UP2 (DW2) is an increase (decrease) and zero otherwise. (t-values are in parentheses) (1) (2) Carve-outs with US parents Cross border carve-outs N 364 76 Mean 0.1955 0.0893 (Median) (0.0712) (0.0516) OVER -0.0007-0.0163 (-0.32) (-2.14) ** LSIZ -0.0276 0.0040 (-1.86) * (0.29) USE 0.0956 0.0870 (1.46) (1.63) DEBT -0.0073-0.0673 (-0.11) (-1.25) VXO 0.9014 0.7993 (2.68)*** (2.66) *** UP1 1.3660 0.0340 (9.57) *** (0.17) DW1 0.2144 0.6367 (0.83) (2.12) ** UP2 1.8678 0.2910 - (6.71) *** (1.20) DW2 0.5095 0.2536 (1.83) * (1.18) F-value 23.49 *** 3.10 *** Adjusted R 2 0.3579 0.2016 Significance at the 10%, 5%, and 1% levels is indicated by 1, 2, and 3 asterisks, respectively. 9

Table 4: Single Factor Analysis Three-Year Returns This table shows three-year market adjusted returns for the total period (1988-2006) for subsidiaries with US parents (column 1); cross border carve-outs (US traded subsidiaries with foreign parents) (column 2); and t-statistics for differences between variables (column 3). Returns are adjusted by subtracting the contemporaneous returns for the CRSP equal-weighted index. The market-adjusted returns are regressed separately on several independent variables: Overhang (OVER), RET/(1-RET); the log of offering size (LSIZ); the proportion of the offering proceeds reinvested by the company (USE); the proportion of the offering as a secondary offering (SEC); the CBOE Volatility Index (VXO); and the ratio of the mid-point of the adjusted filing range to the initial filing range where UP1 (DW1) is positive (negative) and zero otherwise; the ratio of the offering price to the mid-point of the final filing range where UP2 is an increase and zero otherwise; and a control variable for the Sarbanes-Oxley Act where 1 is a 2002 or later carve-out and 0 otherwise. (t-values are in parentheses) (1) (2) (3) T-statistics for Carve-outs with US parents Cross border carve-outs Differences N 364 76 Mean -0.0078-0.0062-1.65 * (Median) (-0.0061) (-0.0052) OVER 0.0001-0.0001 4.29 *** (1.18) (-0.17) LSIZ 0.0000 0.0001-0.76 (0.04) (0.15) USE -0.0033-0.0035 0.75 (-2.42) ** (-1.77) * SEC 0.0030 0.0056-1.86 ** (1.72) * (2.68) *** VXO -0.0301-0.0025 1.02 (-3.15)** (-0.18) UP1-0.0118 0.0093 1.66 ** (-2.88) *** (0.98) DW1 0.0027 0.0149-1.54 * (0.37) (1.07) UP2-0.0130-0.0137 0.40 (-1.76) * (-1.33) SOX 0.0009 0.0056 3.09 *** (0.77) (2.74) *** Significance at the 10%, 5%, and 1% levels is indicated by 1, 2, and 3 asterisks, respectively. Implying low risk for carve-outs, F-values and the negative initial return multiple regression VXO coefficient are significant at the 1% level for carve-outs with US parents. Supporting the partial price adjustment hypothesis of Bradley and Jordan (2002), the positive UP2 parameter is significant at the 1% level for carve-outs with US parents (Table 3, columns 1). The independent variables explain 35.79% and 20.16% of multiple regression initial returns for carve-outs with 10

US parents (Table 3, column 1) and for cross border carve-outs (Table 3, column 2), respectively. Indicating a negative relation between overhang and initial returns and consistent with Bradley and Jordan (2002), cross border carve-outs have negative OVER and positive DW1 coefficients significant at the 5% level (Table 3, column 2). Showing low risk, the negative LSIZ coefficient is marginally significant for carve-outs with US traded parents (Table 3, column 1). For carve-outs with US traded parents (Table 3, column 1) the positive DW2 parameter is marginally significant. 4E. Single variable regressions- three-year returns Table 4 provides single factor three-year market adjusted returns for the period from1988 to 2006. Although not shown in the tables, the difference in initial returns and three-year returns for both carve-outs with US traded parents and for cross border carve-outs are significant at the 1% level. The reversal of positive initial returns to negative long-term market-adjusted returns supports the leaning against the wind hypothesis. The negative USE coefficient is significant at the 5% level or higher for carve-outs with US traded parents (Table 4, column 1) and marginally significant for cross border carve-outs (Table 4, column 2). These results support the managerial hypothesis of Allen and McConnell (1998). Reflecting the inverse of returns for the three-year post carve-out period and consistent with the leaning against the wind hypothesis of Loughran and Ritter (2002), the negative UP1 parameter is significant at the 1% level in Table 4, column 1. The positive SEC coefficient is significant at the 5% level for cross border carve-outs (Table 4, column 2) and marginally significant for carve-outs with US parents (Table 4, column 1). For carve-outs with US parents (Table 4, column 1) the negative UP2 coefficient is marginally significant. These results indicate that the negative three-year returns vary with the SEC and inversely with the UP2 parameters. For carve-outs with US parents, the negative VXO parameter is significant at the 5% level (Table 4, column 1). For cross border carve-outs, the SOX coefficient is significant at the 1% level (Table 4, column 2). Table 4, column 3 reports t-statistics for differences between variables for carve-outs with US parents and for cross border carve-outs. Reflecting low cross border carve-out three-year returns, the -0.16% mean returns are marginally significant.. 4F. Multivariate analysis three-year returns As we did for initial return variables in Section 4.3, we apply the AIC to select our multiple regression variables. The procedure yields models with SEC, OVER, REP, VIX, UP1, UP2, and SOX parameters. Since all VIF are close to 1, the results indicate low multicollinearity. Table 5 reports three-year multiple regression results for the full period (1988-2006). Columns 1and 2 report results for carve-outs with US traded parents and for cross border carve-outs, respectively. 11

Table 5: Multiple Regression Results Three-Year Returns This table shows multiple regressions of three-year market adjusted returns for the period 1988-2006 for subsidiaries with US parents (column 1) and cross border carve-outs (US traded subsidiaries with foreign parents) (column 2). Returns are adjusted by subtracting the contemporaneous returns for the CRSP equal-weighted index. The marketadjusted returns are regressed on several independent variables: the proportion of the offering as a secondary offering (SEC); the ratio of the mid-point of the adjusted filing range to the initial filing range where UP1 is positive and zero otherwise; the ratio of the offering price to the mid-point of the final filing range where UP2 is an increase and zero otherwise; and a control variable for the Sarbanes-Oxley Act where 1 is a 2002 or later carve-out and 0 otherwise. (t-values are in parentheses) (1) (2) Carve-outs with US parents Cross border carve-outs N 364 76 Mean -0.0078-0.0062 (Median) (-0.0061) (-0.0052) SEC 0.0026 0.0050 (1.54) (2.34) ** OVER 0.0001 0.0001 (1.33) (0.18) REP 0.0008 0.0001 (2.13) ** (0.10) VIX -0.0290 0.0063 (-3.01) *** (0.43) UP1-0.0110 0.0018 (-2.69) *** (0.19) UP2-0.0110-0.0094 (-1.50) (-0.94) SOX 0.0004 0.0052 (0.36) (2.45) ** F-value 4.33 *** 2.16 ** Adjusted R 2 0.0524 0.0978 Significance at the 10%, 5%, and 1% levels is indicated by 1, 2, and 3 asterisks, respectively. For cross border carve-outs, the positive SEC coefficient is significant at the 5% level (Table 5, column 2). These results contrast with the negative coefficients for initial returns. Consistent with Beatty and Welch (1989) the positive REP parameter is significant at the 5% level for carve-outs with US traded parents (Table 5, column 1) and indicates that investment bankers redeemed their reputation over the three-year post carve-out period. Reflecting low risk, the negative VIX parameter and F-values are significant at the 1% level for carve-outs with US traded parents (Table 5, column 1). Indicating a positive impact on three-year returns, the positive SOX coefficient is significant at the 5% level for cross border carve-outs (Table 5, column 2). These results are similar to the single variable results in Table 4. Collectively, the 12

independent variables for carve-outs with US parents (Table 5, column 1) explain 5.24% of the variation of three-year returns and those for cross border carve-outs (Table 5, column 2) explain 9.78 %. 5. Conclusions and implications We observe that cross border carve-outs have larger offering sizes, lower initial returns, and higher three-year returns than for other carve-outs and for IPOs. This implies lower risk for cross border carve-outs. Several significant variables (SEC, VXO, UP1, UP2, and SOX), available prior to equity carveout offerings, maintain their significance for the three-year post-carve-out period. For cross border carve-outs, the independent variables explain 20.16% of initial period multiple regression returns. This lessens the need for variables that rely upon information asymmetry. Contrary to Vijh (1999) who found negative, but insignificant three-year returns, we observe negative and significant carve-out returns for a similar period. Also, the difference between initial and three-year returns is significant at the 1% level. These results support the leaning against the wind hypothesis of Loughran and Ritter (2002). 13

References Allen, J.W. and J.J. McConnell, 1998, Equity carve-outs and managerial discretion, The Journal of Finance, 13:1, 163-186. Asquith, P. and D. W. Mullins, Jr., 1986, Equity issues and offering dilution, Journal of Financial Economics, 15, 61-89. Beatty, R.P. and I. Welch, 1996, Issuer expenses and legal liability in initial public offerings, Journal of Law and Economics, 36, 545-602. Bradley, D. J. and B.D. Jordan, 2002, Partial adjustment to public information and IPO underpricing, The Journal of Financial and Quantitative Analysis, 37:4, 593-616. Carter, R.B., F.H. Dark, and A. Singh, 1998, Underwriter reputation, initial returns, and the longrun performance of IPO stocks, The Journal of Finance, 53, 285-312. Dennis, P., S. Mayhew, and C. Stivers, 2006, Stock returns, implied volatility innovations, and the asymmetric volatility phenomenon, Journal of Financial and Quantitative Analysis, 41:2, 381-406. Hanley, K.W. (1993), The underpricing of initial public offerings and the partial adjustment phenomenon, Journal of Financial Economics, 34, 231-20. Hulbert, H.M., J.A. Miles, and J.R. Woolridge, 2002, Value creation from equity carve-outs, Financial Management, 31:1, 83-100. Klein, A., J. Rosenfeld, and W. Beranek, 1991, The two stages of an equity carve-out and the price response of parent and subsidiary stock, Managerial and Decision Economics, 12, 449-460. Leland, H.E. and D.H. Pyle, 1977, Informational asymmetries, financial structure, and financial intermediation, The Journal of Finance, 32:2, 371-387. Logue, D.E., 1973,On the pricing of unseasoned equity issues: 1965-1969, The Journal of Financial and Quantitative Analysis, 8:1, 91-103. Loughran, T. and J.R. Ritter, 1995, The new issues puzzle, The Journal of Finance, 50:1, 23-51. Loughran, T. and J.R. Ritter, 2002, Why don t issuers get upset about leaving money on the table in IPOs? The Review of Financial Studies, 15:2, 413-443. Loughran, T. and J.R. Ritter, 2004, Why has IPO under-pricing changed over time? Financial Management, 35:3, 23-51. Mikkelson, W.H. and M.M. Partch, 1986, Valuation effects of security offerings and the issuance process, Journal of Financial Economics, 15, 31-60. 14

Mulherin, J.H. and A.L. Boone, 2000, Comparing acquisitions and divestitures, Journal of Corporate Finance, 6, 117-139. Ritter, J.R., 1991, The long-run performance of initial public offerings, The Journal of Finance, 46:1, 3-27. Ritter, J.R., 2007, Some factoids about the 2006 ipo market, University of Florida, http://bear.cba.ufl.edu/ritter. Sarbanes-Oxley Act of 2002, 15 US Code (2002). Schipper, K. and A. Smith, 1986, A comparison of equity carve-outs and seasoned equity offerings share price effects and corporate restructuring, Journal of Financial Economics, 15, 153-186. Vijh, A.M., 1999, Long-term returns from equity carve-outs, Journal of Financial Economics, 51, 273-308. Vijh, A.M., 2002, The positive announcement-period returns of equity carve-outs: asymmetric information or divestiture gains? Journal of Business, 75:1, 153-190. 15